Autonomous Learning from the Environment

Wei-Min Shen

Computer Science Press, W.H. Freeman and Company
March 1994, 355 pp.
ISBN 0-7167-8265-0

Foreword by Herbert A. Simon

Table of Contents

Where to Buy


As the fields of Artificial Intelligence and Robotics are being revolutionized by advances in autonomous learning, a conceptual framework for addressing the general problem of learning from the environment has been sorely lacking --- until now. AUTONOMOUS LEARNING FROM THE ENVIRONMENT provides a principled model that integrates once disparate aspects of this burgeoning field and facilitates systematic analysis of present and future autonomous learning systems.

This unique work presents a unified conception of environmental model abstraction, as well as the scientific foundations and practical applications of autonomous learning systems. Drawing on principles of human cognition, it proposes a percept and action based mechanism that enables autonomous systems to continually analyze and adapt to their environment in achieving their goals, hence improving their performance without reprogramming by human intervention.

Combining state-of-the-art theory with illustrative examples and applications, AUTONOMOUS LEARNING FROM THE ENVIRONMENT provides clear, up-to-date coverage of three essential tasks of developing an autonomous learning system: active model abstraction, modal application, and integration. It also contains an implemented system LIVE that autonomously interacts with its environment to solve problems and discover new concepts; readers may use many algorithms in the book to generate their own simulations.

A pathbreaking addition to the professional literature, AUTONOMOUS LEARNING FROM THE ENVIRONMENT is an ideal supplementary text for Artificial Intelligence and Robotics courses and an essential book for any computer science library. As Herbert A. Simon states in the foreword: ``Most readers, I think, will experience more than one surprise as they explore the pages of this book, ...... [it] has provided us with an indispensable vade mecum for our explorations of systems that learn from their environments.''